Update tree-sitter to 0.20

0.20 includes querying improvements, we no longer have to convert
fragments to strings but can return an iterator of chunks instead.

helix-core/Cargo.toml

@@ -22,7 +22,7 @@ unicode-segmentation = "1.8"
 unicode-width = "0.1"
 unicode-general-category = "0.4"
 # slab = "0.4.2"
-tree-sitter = "0.19"
+tree-sitter = "0.20"
 once_cell = "1.8"
 arc-swap = "1"
 regex = "1"

helix-core/src/syntax.rs

@@ -276,16 +276,6 @@ fn byte_range_to_str(range: std::ops::Range<usize>, source: RopeSlice) -> Cow<st
     Cow::from(source.slice(start_char..end_char))
 }
 
-fn node_to_bytes<'a>(node: Node, source: RopeSlice<'a>) -> Cow<'a, [u8]> {
-    let start_char = source.byte_to_char(node.start_byte());
-    let end_char = source.byte_to_char(node.end_byte());
-    let fragment = source.slice(start_char..end_char);
-    match fragment.as_str() {
-        Some(fragment) => Cow::Borrowed(fragment.as_bytes()),
-        None => Cow::Owned(String::from(fragment).into_bytes()),
-    }
-}
-
 impl Syntax {
     // buffer, grammar, config, grammars, sync_timeout?
     pub fn new(
@@ -380,14 +370,11 @@ impl Syntax {
 
         // TODO: if reusing cursors this might need resetting
         if let Some(range) = &range {
-            cursor_ref.set_byte_range(range.start, range.end);
+            cursor_ref.set_byte_range(range.clone());
         }
 
         let captures = cursor_ref
-            .captures(query_ref, tree_ref.root_node(), move |n: Node| {
-                // &source[n.byte_range()]
-                node_to_bytes(n, source)
-            })
+            .captures(query_ref, tree_ref.root_node(), RopeProvider(source))
             .peekable();
 
         // manually craft the root layer based on the existing tree
@@ -501,10 +488,7 @@ impl LanguageLayer {
             //     let mut injections_by_pattern_index =
             //         vec![(None, Vec::new(), false); combined_injections_query.pattern_count()];
             //     let matches =
-            //         cursor.matches(combined_injections_query, tree.root_node(), |n: Node| {
-            //             // &source[n.byte_range()]
-            //             node_to_bytes(n, source)
-            //         });
+            //         cursor.matches(combined_injections_query, tree.root_node(), RopeProvider(source));
             //     for mat in matches {
             //         let entry = &mut injections_by_pattern_index[mat.pattern_index];
             //         let (language_name, content_node, include_children) =
@@ -716,7 +700,7 @@ use std::sync::atomic::{AtomicUsize, Ordering};
 use std::{iter, mem, ops, str, usize};
 use tree_sitter::{
     Language as Grammar, Node, Parser, Point, Query, QueryCaptures, QueryCursor, QueryError,
-    QueryMatch, Range, Tree,
+    QueryMatch, Range, TextProvider, Tree,
 };
 
 const CANCELLATION_CHECK_INTERVAL: usize = 100;
@@ -776,32 +760,57 @@ struct LocalScope<'a> {
 }
 
 #[derive(Debug)]
-struct HighlightIter<'a, 'tree: 'a, F>
+struct HighlightIter<'a, F>
 where
     F: FnMut(&str) -> Option<&'a HighlightConfiguration> + 'a,
 {
     source: RopeSlice<'a>,
     byte_offset: usize,
     injection_callback: F,
     cancellation_flag: Option<&'a AtomicUsize>,
-    layers: Vec<HighlightIterLayer<'a, 'tree>>,
+    layers: Vec<HighlightIterLayer<'a>>,
     iter_count: usize,
     next_event: Option<HighlightEvent>,
     last_highlight_range: Option<(usize, usize, usize)>,
 }
 
-struct HighlightIterLayer<'a, 'tree: 'a> {
+// Adapter to convert rope chunks to bytes
+struct ChunksBytes<'a> {
+    chunks: ropey::iter::Chunks<'a>,
+}
+impl<'a> Iterator for ChunksBytes<'a> {
+    type Item = &'a [u8];
+    fn next(&mut self) -> Option<Self::Item> {
+        self.chunks.next().map(str::as_bytes)
+    }
+}
+
+struct RopeProvider<'a>(RopeSlice<'a>);
+impl<'a> TextProvider<'a> for RopeProvider<'a> {
+    type I = ChunksBytes<'a>;
+
+    fn text(&mut self, node: Node) -> Self::I {
+        let start_char = self.0.byte_to_char(node.start_byte());
+        let end_char = self.0.byte_to_char(node.end_byte());
+        let fragment = self.0.slice(start_char..end_char);
+        ChunksBytes {
+            chunks: fragment.chunks(),
+        }
+    }
+}
+
+struct HighlightIterLayer<'a> {
     _tree: Option<Tree>,
     cursor: QueryCursor,
-    captures: iter::Peekable<QueryCaptures<'a, 'tree, Cow<'a, [u8]>>>,
+    captures: iter::Peekable<QueryCaptures<'a, 'a, RopeProvider<'a>>>,
     config: &'a HighlightConfiguration,
     highlight_end_stack: Vec<usize>,
     scope_stack: Vec<LocalScope<'a>>,
     ranges: Vec<Range>,
     depth: usize,
 }
 
-impl<'a, 'tree: 'a> fmt::Debug for HighlightIterLayer<'a, 'tree> {
+impl<'a> fmt::Debug for HighlightIterLayer<'a> {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         f.debug_struct("HighlightIterLayer").finish()
     }
@@ -972,7 +981,7 @@ impl HighlightConfiguration {
     }
 }
 
-impl<'a, 'tree: 'a> HighlightIterLayer<'a, 'tree> {
+impl<'a> HighlightIterLayer<'a> {
     /// Create a new 'layer' of highlighting for this document.
     ///
     /// In the even that the new layer contains "combined injections" (injections where multiple
@@ -1029,10 +1038,7 @@ impl<'a, 'tree: 'a> HighlightIterLayer<'a, 'tree> {
                         let matches = cursor.matches(
                             combined_injections_query,
                             tree.root_node(),
-                            |n: Node| {
-                                // &source[n.byte_range()]
-                                node_to_bytes(n, source)
-                            },
+                            RopeProvider(source),
                         );
                         for mat in matches {
                             let entry = &mut injections_by_pattern_index[mat.pattern_index];
@@ -1079,10 +1085,7 @@ impl<'a, 'tree: 'a> HighlightIterLayer<'a, 'tree> {
                     let cursor_ref =
                         unsafe { mem::transmute::<_, &'static mut QueryCursor>(&mut cursor) };
                     let captures = cursor_ref
-                        .captures(&config.query, tree_ref.root_node(), move |n: Node| {
-                            // &source[n.byte_range()]
-                            node_to_bytes(n, source)
-                        })
+                        .captures(&config.query, tree_ref.root_node(), RopeProvider(source))
                         .peekable();
 
                     result.push(HighlightIterLayer {
@@ -1236,7 +1239,7 @@ impl<'a, 'tree: 'a> HighlightIterLayer<'a, 'tree> {
     }
 }
 
-impl<'a, 'tree: 'a, F> HighlightIter<'a, 'tree, F>
+impl<'a, F> HighlightIter<'a, F>
 where
     F: FnMut(&str) -> Option<&'a HighlightConfiguration> + 'a,
 {
@@ -1287,7 +1290,7 @@ where
         }
     }
 
-    fn insert_layer(&mut self, mut layer: HighlightIterLayer<'a, 'tree>) {
+    fn insert_layer(&mut self, mut layer: HighlightIterLayer<'a>) {
         if let Some(sort_key) = layer.sort_key() {
             let mut i = 1;
             while i < self.layers.len() {
@@ -1306,7 +1309,7 @@ where
     }
 }
 
-impl<'a, 'tree: 'a, F> Iterator for HighlightIter<'a, 'tree, F>
+impl<'a, F> Iterator for HighlightIter<'a, F>
 where
     F: FnMut(&str) -> Option<&'a HighlightConfiguration> + 'a,
 {
@@ -1570,7 +1573,7 @@ where
 fn injection_for_match<'a>(
     config: &HighlightConfiguration,
     query: &'a Query,
-    query_match: &QueryMatch<'a>,
+    query_match: &QueryMatch<'a, 'a>,
     source: RopeSlice<'a>,
 ) -> (Option<Cow<'a, str>>, Option<Node<'a>>, bool) {
     let content_capture_index = config.injection_content_capture_index;

helix-syntax/Cargo.toml

@@ -11,7 +11,7 @@ homepage = "https://helix-editor.com"
 include = ["src/**/*", "languages/**/*", "build.rs", "!**/docs/**/*", "!**/test/**/*", "!**/examples/**/*", "!**/build/**/*"]
 
 [dependencies]
-tree-sitter = "0.19"
+tree-sitter = "0.20"
 libloading = "0.7"
 anyhow = "1"